The subject application is directed generally to improving quality of digitally encoded color images. The application is particularly applicable to isolation or detection of facial areas in image data and improving the color characteristics thereof.
More recently, images are acquired as encoded, digital image data. Such image data is obtained via devices such as digital cameras, both for still images and moving pictures. Digital image data is also acquired by scanning of tangible images, such as from pictures or negatives obtained via more traditional, film-based cameras. Color image data is typically encoded in a multidimensional color space, such as red-green-blue (RGB); cyan-magenta-yellow (CYM), which may include black (CYMK); or any other of a plurality of alternative encoding schemes.
A large number of images include depictions of human faces, which frequently form the area or areas of greatest interest to a viewer. It is desirable to have as accurate a detection of facial images as possible. Earlier systems, operable on a black-and-white rendering of a digital image, sought to isolate a facial region for special treatment. However, such systems are prone to false positives, and are generally separated from any system that seeks to perform image correction.